Many-objective day-ahead optimal scheduling of residential flexible loads integrated with stochastic occupant behavior models

•Feature parameters were determined to describe occupant behaviors for load scheduling.•A general and easy-to-use stochastic occupant behavior model was developed.•Many-objective day-ahead optimal scheduling of residential flexible loads.•Daily electricity costs are decreased by 6.25 % by using the optimal plan. Residential flexible load scheduling is playing a more and more vital role in balancing the supply and demand of utility grids with the increasing penetration of solar and wind power generation. Nevertheless, the optimal scheduling of residential flexible loads is significantly impacted by occupant behaviors of household appliance usage. In previous studies, occupant behaviors were generally oversimplified with pre-defined schedules, which ultimately renders optimal dispatch plans infeasible. To address this issue, a new framework has been proposed to optimize the day-ahead scheduling of residential flexible loads, which considers the stochastic occupant behaviors. Firstly, the feature parameters to characterize occupant behaviors of thermostatically controlled loads and deferrable loads were proposed, considering the operation characteristics and scheduling optimization of these flexible loads. A general and easy-to-use stochastic occupant behavior model was then developed using Metropolis Hasting algorithm-based Markov chain Monte Carlo method. The feature parameters derived from the validated stochastic occupant behavior model were then input into the established many-objective optimal model to perform day-ahead optimal dispatch, considering the benefits of various stakeholders. NSGA-III and TOPSIS methods were employed to solve the optimal model. The proposed framework was found to be feasible, as the daily electricity costs, the daily CO2 emissions, and the average ramping index of household power profiles were all decreased by 6.25 %, 6.3 %, and 15.0 %, respectively, when compared to the benchmark. Overall, this framework provides practical guidance for more accurate and reliable day-ahead optimal scheduling of residential flexible loads.